Together with the ten species that have been delisted due to extinction, 81 (5%) current or formerly ESA-listed species are thought, to some greater or lesser extent, to be extinct.
Extinction and the ESA
Abstract
The success of the Endangered Species Act is often measured by comparing the number of species declared extinct to the number of listed species, which indicates success on the order of 99%. Here we check that measure against the discussions of extinction by the Fish and Wildlife Service and the National Marine Fisheries Service by computationally mining five-year status reviews. We find that 81 (~5%) ESA-listed species are known to be or are likely to be extinct. The extinctions are concentrated in Hawaii and a swath that extends from Mississippi to Virginia.
The extinction crisis of the modern era was a key motvation for the ESA: the goals of the law include preventing extinction and recovering species so they are no longer threatened with extinction. One of the ways ESA success is measured is by comparing the number of listed species to the number that are extinct. As of this writing, ten species have been removed from the list of threatened and endangered species, and the list includes 1,652 species. Those numbers suggest 99.4% of species have been saved by the ESA, which is a great accomplishment.
But there are a few problems with using the number of species “delisted due to extinction” as the metric of success. First, it ignores the key goal of species recovery and uses the lower bar of mere existence as success. But we don’t want to argue that plants and animals known only to exist in captivity count as successes. Success is wildlife out in their homes on the landscape.
Second, we know that the ten delisted species are simply the ones formally accepted as extinct. But proving extinction is tough; we have too many examples where a species hadn’t been seen for many, many years–multiple decades and even over a century–only to be rediscovered. For such reasons, declarations of extinction are a rare occurrence. At the same time, we often have a very good idea that well-documented species are really-and-truly extinct. We hold off on the declarations of extinction and the removal of protections because of the small but real chance that one or a few individuals still exist.
Here we identify the species that experts - biologists with the U.S. Fish and Wildlife Service (FWS) and National Marine Fisheries Service (NMFS; collectively, the Services) - have good reason to believe are or may be extinct. Declaring extinction may not be warranted for many or most of these, but these professionals have indicated that extinction is likely to some greater or lesser degree. Getting a better handle on the number of species thought to be extinct rather than relying on the number delisted due to extinction is an important first step in thoroughly understanding the effectiveness of the ESA.
Identifying possible/probable extinctions
The first question we address is how many species are extinct. Pinning down that number is tricky because extinction is fundamentally hard: one can’t prove absence the same way we prove presence. Rather than a declaration of extinct, we set the bar at a level of experts’ statements that they think the species is or may be extinct. We mined the text of all ~1,400 five-year status reviews, totalling >30,000 pages, we need to find a relatively small number of phrases that indicate extinction.
This search returned 387 matches to the set of candidate phrase patterns, including patterns that indicate a higher likelihood of extinction such as:
|
|
|
|
|
|
|
|
|
|
Perusing the pattern matches, we find (as expected) both true positives (i.e., the meaning of the matched sentence comports with the idea we have in mind) and false positives (e.g., local extinctions or species previously believed extinct but since re-discovered). To ensure we are working with the right numbers, we manually checked each 5-year review with an extinction phrase match to remove the false-positives.
From the manually filtered data we find 37 species that are believed extinct to some greater or lesser degree, plus the snail genus Achatinella (see below).
| Acaena exigua |
| Achatinella spp. |
| Akialoa stejnegeri |
| Amaranthus brownii |
| Anolis roosevelti |
| Cyanea copelandii ssp. copelandii |
| Cyanea superba |
| Cyrtandra crenata |
| Dipodomys heermanni morroensis |
| Eleutherodactylus jasperi |
| Epioblasma torulosa torulosa |
| Erimystax cahni |
| Etheostoma sellare |
| Hemignathus hanapepe |
| Hibiscadelphus woodii |
| Kadua degeneri |
| Loxops ochraceus |
| Melamprosops phaeosoma |
| Melicope quadrangularis |
| Moho braccatus |
| Myadestes lanaiensis rutha |
| Myadestes myadestinus |
| Numenius borealis |
| Obovaria retusa |
| Ochrosia kilaueaensis |
| Paroreomyza flammea |
| Paroreomyza maculata |
| Phyllostegia glabra var. lanaiensis |
| Phyllostegia mollis |
| Pleurobema curtum |
| Pleurobema marshalli |
| Polygyriscus virginianus |
| Psittirostra psittacea |
| Pteropus tokudae |
| Puma (=Felis) concolor couguar |
| Quadrula stapes |
| Tetramolopium capillare |
| Zosterops conspicillatus conspicillatus |
(Note this does not include species that are extinct in the wild, but often number just a few in propagation or captivity. We identified 6 such species in our scans - Delissea rhytidosperma, Hibiscadelphus giffardianus, Phyllostegia kaalaensis, Cyanea superba, Rallus owstoni, Hibiscadelphus hualalaiensis - but the number is likely higher.)
Most of the extinct species are from Hawaii, with many that are plants Box 1 or snails Box 2. Note that we collapse all 31 extinct species in the genus Achatinella to “Achatinella spp.” in subsequent analyses because they were all from the same place and faced the same threats. To count them individually when trying to understand broad patterns related to extinction would bias our inferences.
Geography of extinctions
The next question is where the extinctions have occurred. Hawaii, of course, tops the list, but where are the other species?
States with the most (likely) extinctions.
Again, Hawaii clearly shows up with the most species, but the swath from Mississippi to Virginia stands out too. Most of these species are mussels and fish, so the issue could be related to the problems apparent in this (albeit old) Clean Water Act 303(d) impaired waters map:
Impairment as measured by EPA standards is not the complete answer why are extinctions not higher in other states with realtively high impairment.
| state | n |
|---|---|
| Hawaii | 24 |
| Alabama | 4 |
| Mississippi | 4 |
| Guam | 3 |
| Tennessee | 3 |
| Virginia | 3 |
In addition to the state-level patterns of extinction, we can use counties of occurrence to evaluate patterns of the geography of extinction. The ten counties with the most extinct species include:
| State, County | n |
|---|---|
| Honolulu, Hawaii | 8 |
| Kauai, Hawaii | 7 |
| Maui, Hawaii | 7 |
| Hawaii, Hawaii | 5 |
| Guam, Guam | 3 |
| Bedford, Tennessee | 2 |
| Bland, Virginia | 2 |
| Blount, Tennessee | 2 |
| Buchanan, Virginia | 2 |
| Butler, Kentucky | 2 |
As expected, Hawaiian counties top the list, followed by Guam, then southern Appalachian counties in Tennessee, Virginia, and Kentucky. Thus, although Alabama and Mississippi have more extinct, listed species than these states, those species are (were) more concentrated in Tennessee, Virginia, and Kentucky.
Common threats and habitats
Given the list of likely extinct species, we would like to know what they have in common. Any commonalities might help inform prioritization efforts that can stem future extinctions. We build off of the data available through the IUCN’s Red List database.
Habitats
| habitat | n |
|---|---|
| NA | 21 |
| Forest - Subtropical/Tropical Moist Montane | 11 |
| Forest - Subtropical/Tropical Moist Lowland | 8 |
| Forest - Subtropical/Tropical Dry | 6 |
| Wetlands (inland) - Permanent Rivers/Streams/Creeks (includes waterfalls) | 3 |
| Rocky areas (eg. inland cliffs, mountain peaks) | 2 |
| Grassland - Subtropical/Tropical Dry | 1 |
| Grassland - Tundra | 1 |
| Marine Coastal/Supratidal - Sea Cliffs and Rocky Offshore Islands | 1 |
| Wetlands (inland) - Bogs, Marshes, Swamps, Fens, Peatlands | 1 |
The most common defined habitats for the extinct species parallels those species’ geography: subtropical / tropical forests of Hawaii where species such as the Achatinella snails, various plants, and birds lived. Inland wetlands, in particular the rivers, streams, and creeks inhabited by likely extinct mussels and fishes, are the second major group. Last, notice the most common habitat type is actually “NA.” This points to a limitation of the IUCN’s database - incomplete information - that we will propose to help address in another venue.
Threats
| title | n |
|---|---|
| Invasive non-native/alien species/diseases | 22 |
| Unspecified species | 22 |
| Invasive non-native/alien species/diseases (Sus domesticus) | 14 |
| Named species (Sus domesticus) | 14 |
| NA | 14 |
| Named species | 11 |
| Livestock farming & ranching | 9 |
| Logging & wood harvesting | 7 |
| Annual & perennial non-timber crops | 6 |
| Fire & fire suppression | 6 |
| code | n |
|---|---|
| 8.1 | 90 |
| 8.1.2 | 74 |
| 8.1.1 | 20 |
| NA | 14 |
| 2.3 | 9 |
| 5.3 | 7 |
| 2.1 | 6 |
| 5.3.3 | 6 |
| 7.1 | 6 |
| 11.1 | 5 |
Box 1. Complementary data from PEP
The Plant Extinction Prevention (PEP) Program works exhaustively to monitor and conserve Hawaiian plants. They also provide more up- to-date data on the status of many imperiled plants than FWS documents, and should probably be considered authoritative for PEP species.
We downloaded the 2017-03-16 list from PEP’s website then used Adobe’s online extraction service to render the tables as a spreadsheet for processing. To begin, we have a high-level overview of how PEP classifies the covered species:
| PEP_class | n |
|---|---|
| N/A | 1 |
| PEP | 227 |
| PEP-Extinct in Wild | 14 |
| PEP-EXTINCT? | 20 |
| PEP-EXTIRPATED | 45 |
| PEP-Extirpated in Wild | 3 |
| PEP-EXTIRPATED? | 14 |
Appropriately, PEP uses “?” to denote species for which extirpation and extinction are unknown. The refinement of extinct in the wild vs. extinct from the planet is also useful to consider. Last, in addition to the extinctions, the PEP data highlight the problem of extirpations: 34 PEP species have been (or probably have been) extirpated from one or more of the islands. First, the number of extirpations by island:
| extirpations | |
|---|---|
| Maui | 14 |
| Molokai | 14 |
| Lanai | 10 |
| Kauai | 7 |
| Oahu | 5 |
| Hawaii | 4 |
| Niihau | 4 |
| Kahoolawe | 1 |
And second, by most species:
| extirpations | |
|---|---|
| Scaevola coriacea | 5 |
| Adenophorus periens | 4 |
| Diplazium molokaiense | 4 |
| Isodendrion pyrifolium | 4 |
| Canavalia pubescens | 3 |
| Acaena exigua | 2 |
| Asplenium dielerectum | 2 |
| Brighamia rockii | 2 |
| Deparia kaalaana | 2 |
| Gardenia brighamii | 2 |
The PEP data adds depth to our understanding of extinctions and extirpations among Hawaiian plants, but PEP doesn’t cover all ESA-listed plants. By combining PEP table with county occurrence data, we find 320 species are not covered by PEP; more are missing than are covered by PEP. For these species in particular it is important for FWS to track species status. One way that is done is with ESA-mandated five-year reviews of species status.
There are 279 non-PEP species with 5-year reviews available, and of those reviews, 200 are out-of-date (i.e., more than five years old).
Box 2. Achatinella snails
Some 43 species of Achatinella snails have been described, all native to the tropical forests of Oahu. Unfortunately, they have suffered more likely extinctions than any other genus of ESA-listed species. Here we summarize the status as given in the last five-year review for the genus (2011) plus status information gleaned from other sources (e.g., IUCN and NatureServe).
Among the 43 Achatinella, current publicly available information indicates 33 are extinct:
| species | year_most_recent_info | year_last_observed |
|---|---|---|
| Achatinella abbreviata | 2009 | 1963 |
| Achatinella apexfulva_cestus | 2009 | 1966 |
| Achatinella apexfulva_leucorraphe | 2011 | 1989 |
| Achatinella apexfulva_vittata | 2011 | 1953 |
| Achatinella bellula | 2011 | 1981 |
| Achatinella buddii | 2011 | 1900 |
| Achatinella bulimoides | 2011 | 2009-2011 |
| Achatinella bulimoides_elegans | 2011 | 1952 |
| Achatinella bulimoides_rosea | 2011 | 1949 |
| Achatinella caesia | 2011 | 1900 |
| Achatinella casta | 2011 | NA |
| Achatinella curta | 2011 | 1989 |
| Achatinella decora | 2011 | 1900 |
| Achatinella dimorpha | 2011 | 1967 |
| Achatinella juddii | 2011 | 1958 |
| Achatinella juncea | 2011 | NA |
| Achatinella lehuiensis | 2011 | 1922 |
| Achatinella lorata | 2011 | 1979 |
| Achatinella papyracea | 2011 | NA |
| Achatinella phaeozona | 2011 | 1974 |
| Achatinella pulcherrima | 2011 | 1993 |
| Achatinella pupukanioe | 2011 | 1980 |
| Achatinella spaldingi | 2011 | 1938 |
| Achatinella stewartii | 2011 | 1963 |
| Achatinella swiftii | 2011 | 1970 |
| Achatinella taeniolata | 2011 | 1966 |
| Achatinella thaanumi | 2011 | 1900 |
| Achatinella turgida | 2011 | 1974 |
| Achatinella valida | 2011 | 1951 |
| Achatinella viridans | 2011 | 1979 |
| Achatinella vulpina | 2011 | 1965 |
Methods
Patterns
We used a set of seven regular expression patterns as a coarse filter:
"(possibly|may be|have become) extinct""(certainly being|probably|is|likely|probably being) extinct""no (populations|individuals)( are)* known""no( known| living)* individuals""extinct in( the)* wild""last (seen|observed|found) in [0-9oOiI]+""functionally extinct"
In these patterns, “|” means “or” for the set of words inside parentheses. For example, the first pattern would match “possibly extinct,” “may be extinct,” or “have become extinct.” The “*" means the preceding letter or word matches zero or more times, whereas “+” means the preceding word/letter(s) must match one or more times. We checked all seven patterns against all 1,385 five-year reviews we collected from ECOS and NMFS’s recovery site.